Publisher's Synopsis
Combat actions are planned based on the best available information. In nearly all situations, significant uncertainty about the combat environment exists. This uncertainty contributes largely to friendly and non-combatant casualties. At the tactical level, operators are often required to enter hostile-occupied buildings without knowledge of the building layout. Military operators have begun to use robots to assist in missions of this type. In general, currently fielded robots lack autonomy and the ability to disseminate an accurate map, on-site, in real time. The purpose of this thesis is to examine the feasibility of an autonomous robot that can localize and build accurate 3D maps using only light detection and ranging (LIDAR). To accomplish this, a robot equipped with only LIDAR and a control algorithm for LIDAR localization and mapping (LLAM) were developed. Trials were then developed to determine if LLAM is a feasible model for interior 3D mapping. Navigation was accomplished using a potential field model adapted from previous work combined with the Hybrid A* search algorithm. Mapping and localization were conducted using the iterative closest point and normal distribution transform methods of point cloud registration. Experimentation revealed that LLAM is a feasible method for interior 3D mapping in real time. Further development of the algorithm may make fielding a LIDAR-equipped mapping robot possible with current mobile computing technology.
